Ductile crack growth—III. Transition to cleavage fracture incorporating statistics

Abstract

The fracture resistance of ferritic steels in the ductile/brittle transition regime is controlled by the competition between ductile tearing and cleavage fracture. Under typical conditions, a crack initiates and grows by ductile tearing but ultimate failure occurs by catastrophic cleavage fracture. In this study the tearing process is simulated using void-containing cell elements embedded within a conventional elastic-plastic continuum; details of the cell model are discussed in Parts I and II of this article. Weakest link statistics is incorporated into the cell element model and this new model is employed to predict the onset of unstable cleavage fracture. Our approach differs from previous analyses in several important ways. The elastic-plastic field computed for crack growth by ductile tearing is fully integrated with a weakest link cleavage fracture model. The model also accounts for the competition between the nucleation of voids from carbide inclusions and the unstable cracking of inclusions precipitating catastrophic cleavage fracture. This model leads immediately to a natural definition of the Weibull stress measure pertinent to cleavage fracture. The model is not restricted by the extent of plastic deformation and ductile tearing. Two effects are associated with ductile crack growth: the cumulative sampling volume is increased and the crack tip constraint is altered. Both effects have important roles which are treated within the present cleavage fracture model. Load-displacement behavior, ductile tearing resistance and transition to cleavage fracture are investigated for several different test geometries and a range of microstructural parameters. It is found that certain variations in microstructure can result in pronounced effects on the cleavage fracture toughness though they have no effect on the ductile tearing resistance preceding cleavage. Rate effects on ductile tearing and transition to cleavage fracture are also discussed. The model predicts trends in ductile/brittle transition that are consistent with available experimental data.